Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 Jul 6.
Published in final edited form as: Am J Infect Control. 2014 Oct;42(10 0):S203–S208. doi: 10.1016/j.ajic.2014.05.020

On the CUSP: Stop BSI: Evaluating the relationship between central line–associated bloodstream infection rate and patient safety climate profile

Sallie J Weaver a,b,*, Kristina Weeks a, Julius Cuong Pham a,c
PMCID: PMC4491500  NIHMSID: NIHMS697450  PMID: 25239711

Abstract

Background

Central line–associated bloodstream infection (CLABSI) remains one of the most common and deadly hospital acquired infections in the United States. Creating a culture of safety is an important part of healthcare–associated infection improvement efforts; however, few studies have robustly examined the role of safety climate in patient safety outcomes. We applied a pattern-based approach to measuring safety climate to investigate the relationship between intensive care unit (ICU) patient safety climate profiles and CLABSI rates.

Methods

Secondary analyses of data collected from 237 adult ICUs participating in the On the CUSP: Stop BSI project. Unit-level baseline scores on the Hospital Survey on Patient Safety, a survey designed to assess patient safety climate, and CLABSI rates, were investigated. Three climate profile characteristics were examined: profile elevation, variability, and shape.

Results

Zero-inflated Poisson analyses suggested an association between the relative incidence of CLABSI and safety climate profile shape. K-means cluster analysis revealed 5 climate profile shapes. ICUs with conflicting climates and nonpunitive climates had a significantly higher CLABSI risk compared with ICUs with generative leadership climates.

Conclusions

Relative CLABSI risk was related to safety climate profile shape. None of the climate profile shapes was related to the odds of reporting zero CLABSI. Our findings support using pattern-based methods for examining safety climate rather than examining the relationships between each narrow dimension of safety climate and broader safety outcomes like CLABSI.

Keywords: Patient safety, Organizational culture, Central line–associated bloodstream infection

Central line–associated bloodstream infection (CLABSI) remains one of the most common and deadly hospital acquired infections in the United States. The most recently available estimates from the Centers for Disease Control and Prevention (CDC) indicate that approximately 41,000 patients experienced a CLABSI in 2011, and that approximately 1 in 4 affected patients died as a result.1 In addition, CLABSIs represent a significant cost burden, with an estimated $17,000 (range $7288–$29,156) in added care expenses associated with each such infections.1,2 Widespread patient safety efforts to reduce CLABSI focus on both the technical aspects of care (eg, clinical care procedures) and adaptive aspects of care (eg, behavioral norms among unit clinicians and staff regarding patient safety, teamwork, and communication, as reflected in the unit’s safety climate)3,4; however, few previous studies have meaningfully examined the relationship between adaptive aspects of care, such as patient safety climate, and patient outcomes in a meaningful way.

Patient safety climate can be defined as the collection of habits, policies, procedures, and behaviors observed in daily practice related to patient safety that are shared among members of a unit, team, or organization.5,6 Specifically, safety climate is a multidimensional concept comprising several aspects, including teamwork and communication among care providers, peer and leadership responses to patient safety events or concerns, and management support for patient safety activities and those who act in the name of patient safety. In this way, the concept of safety climate is like a cake that is composed of multiple ingredients. It is not one single ingredient that makes a cake, but rather the cake emerges from complex interactions and patterns among the various ingredients that have gone into it. In much the same way, safety climate can be thought of as an overarching concept that emerges from all of the different dimensions that compose it.

Safety climate, and the closely related concept of safety culture, have been publicized as critical elements for reducing hospital-acquired infections7–9; however, studies examining the nature of the relationship between safety climate and unit-level CLABSI rates are needed. Previous studies have identified relationships between certain individual dimensions of safety climate and length of hospital stay,10,11 in-hospital complications and adverse events,12 risk-adjusted mortaility,10 and clinician compliance with safe work practices.13 Yet findings vary, and the magnitude of the observed relationships is inconsistent, making it difficult to determine how much patient safety climate matters when it comes to hospital-acquired infections.

A potential alternative explanation for these varied findings is related to the way in which previous studies elected to operationalize the analysis of patient safety climate. Most previous studies took what is termed a “reductionist” approach to examining the safety climate–outcome relationship, meaning that they tested the relationship between each individual dimension of safety climate (ie, each individual ingredient) and a targeted outcome, rather than studying the relationship between safety climate as an entire concept (ie, the cake) and the given outcome of interest. Examining the dimensions individually can be problematic from both a theoretical and an analytical perspective, and can lead to weak observed relationships and conflicting findings.14–17 For example, when the dimensions of safety climate are investigated individually, there is an inherent mismatch between the bandwidth of the predictor (ie, a single dimension measured using a safety climate survey) and the bandwidth of the outcome (eg, infection rates) that makes it more difficult to detect true relationships between safety climate and patient safety outcomes.

To address this gap, the present study drew on configural (ie, pattern-based) theories of organizational culture and climate18,19 to examine the association between the constellation of dimensions that comprise patient safety climate and CLABSI rates in a sample of adult intensive care units (ICUs) in the United States. Configural theories of organizational climate and culture suggest that the pattern of the different dimensions of safety climate can be described in terms of 3 profile characteristics: (1) profile elevation, the general positive or negative valence of the safety climate across all different dimensions; (2) profile variance, the degree of variation among different climate dimensions; and (3) profile shape, the specific pattern of peaks and valleys among the different climate dimensions. We adopted a pattern-based methodology to investigate the relationships among these 3 safety climate profile characteristics and our primary outcome of interest.

METHODS

Secondary analyses were conducted using a subset of data collected as part of the On the CUSP: Stop BSI project, a national improvement collaborative funded by the Agency for Healthcare Research and Quality (AHRQ). The Stop BSI program used a multiple time series design to evaluate the effectiveness a multifaceted intervention that included the Comprehensive Unit-Based Safety Program(CUSP),3,20,21 a model for translating research into practice that includes using a checklist of best practices to prevent infection and providing feedback on infection rates as a strategy to reduce CLABSI rates in ICU settings. Baseline safety climate survey data and baseline bloodstream infection data from the first 4 cohorts of participating adult ICUs (n = 238) were included for secondary analysis. The majority of units (78%) were located in nonrural areas, and 50% of these units were part of a teaching hospital.

Measures

Safety climate

Safety climate was measured using the Hospital Survey on Patient Safety (HSOPS),22 which measures 10 distinct dimensions of safety climate (see Table 1) and has demonstrated sound psychometric properties across a variety of acute care settings.23 Each dimension comprises 3 or 4 questions that are aggregated to form a dimension score for each unit. In line with scoring recommendations from the survey developer, a unit-level score on each of the 10 dimensions is calculated as the average percentage of positive responses (score of 4 or 5 on a 5-point Likert response scale) across all respondents in a given unit. Thus, the “percent positive score” can range from 0% to 100% on each of the 10 dimensions for each unit, with higher scores indicating more positive or desirable climate characteristics.

Table 1.

Unit level climate shape descriptive statistics

Climate shape Teamwork
within ICU		 Supervisor
expectations		 Continuous
learning		 Staffing and
workload		 Nonpunitive
response		 Feedback
about error		 Open
communication		 Hosptial
management
support		 Teamwork
across ICUs		 Handoffs and
care transitions
Generative leadership
  Mean 86.10 78.37 72.46 58.00 41.09 60.71 64.08 59.16 50.72 42.48
  SD 7.29 7.38 6.37 10.10 10.73 9.81 9.27 9.26 7.55 6.87
Nonpunitive
  Mean 90.21 84.10 83.13 67.24 45.95 70.38 69.50 75.81 62.67 52.39
  SD 5.11 6.21 4.90 9.22 7.93 9.43 6.46 7.22 8.62 8.47
Team-oriented
  Mean 76.76 63.47 61.43 46.32 25.40 45.71 52.11 47.77 45.58 37.53
  SD 9.87 10.78 8.21 9.25 8.37 9.75 7.46 10.33 9.68 8.85
Comprehensive
  Mean 95.13 91.72 91.47 79.66 62.34 79.11 79.87 87.60 78.34 69.60
  SD 3.60 5.79 5.17 8.65 14.37 7.20 8.78 6.62 9.90 10.21
Conflicting
  Mean 87.62 71.67 76.28 58.34 30.00 59.37 58.81 65.85 62.98 55.06
  SD 6.49 7.98 6.37 9.41 8.37 9.87 8.52 8.56 6.96 7.47
F test statistic 38.48* 75.77* 127.47* 64.34* 78.61* 72.10* 63.68* 123.85* 81.64* 81.18*
ICCcluster (95% CI) 0.41 (0.16–0.72) 0.59 (0.29–0.84) 0.71 (0.41–0.90) 0.57 (0.27–0.83) 0.64 (0.33–0.86) 0.58 (0.28–0.83) 0.57 (0.27–0.83) 0.71 (0.41–0.90) 0.64 (0.33–0.86) 0.64 (0.33–0.86)
Open in a new tab

ICCcluster: 1, 61 ICUs; 2, 61 ICUs; 3, 54 ICUs; 4, 24 ICUs; 5, 38 ICUs.

NOTE. To determine whether the 5 profile shapes differed meaningfully from one another in terms of safety climate scores, ANOVA was performed on each safety climate dimension. A significant F value indicates that the shapes demonstrated significantly different scores on the particular dimension. In addition, ICCs are reported to show meaningful similarities among respondents categorized with the same climate profile shape.

*

P < .0001.

Participating ICUs had the option to administer the survey to clinicians and staff working in participating ICUs using a Web-based survey platform at the onset of the project or to submit HSOPS data that their unit had already collected during annual safety climate measurement activities within their hospital if collected recently. Project leaders for the ICUs worked in partnership with the national project team to distribute surveys using an online survey platform and to provide background data regarding their unit, including bed size and type. Baseline HSOPS data were collected during the first 30–60 days of each cohort’s participation in the project. Cohort 1 collected baseline HSOPS data between September and November 2009. Cohort 2 and a portion of cohort 1 collected baseline HSOPS data between October and December 2009. Cohort 3 collected baseline HSOPS data between May and June 2010, and cohort 4 collected survey data between September and October 2010.

Operationalizing the climate profile characteristics for each unit

The 3 safety climate profile characteristics (profile elevation, variability, and shape) were operationalized for each unit based on the 10 dimension scores. Profile elevation was computed as a single score for each ICU representing the mean percent positive score across all 10 safety climate dimensions. Profile variability was similarly computed for each ICU as the variance of the 10 dimension scores around their respective mean. In line with previous studies examining organizational climate profiles,19 profile shape was operationalized using k-means cluster analysis, which uses algorithmic iterations to group individual ICUs into relatively homogeneous groups based on a battery of selected characteristics, such as scores on each of the 10 safety climate dimensions. Within health care, similar clustering methods have been used to examine the relationship between health care personnel attitudes about risk and influenza vaccination uptake and absenteeism.24 The method used to derive the 5 different profile shapes that emerged from this cluster analysis are fully described in the next section.

Operationalizing climate shape

During exploratory data analysis, potential k-means cluster solutions for 2–6 clusters were examined. The best-fitting cluster solution was selected based on the Calinski and Harabasz30 criterion, as well as on examination of cluster interclass correlations (ICCs) and theoretical interpretation of resulting clusters. Our results suggested a 5-cluster solution as the best-fitting, most theoretically sound solution. The 5 climate profile shapes are displayed in Figure 1. We drew on other configuration-based theories of organizational climate and culture as theoretical grounding for the 5 shapes that emerged, including the competing values framework.19,31,32 Profile shape 1, generative leadership climate, describes units in which high levels of hospital leadership support for patient safety and collaboration across units and services is perceived as a priority, even relative to teamwork among ICU team members. In these climates, organizational leadership plays a significant role in motivating and reinforcing patient safety as the top organizational priority. Conversely, a nonpunitive climate shape is one in which the peers and unit leaders demonstrate a blame-free response to error and unit members perceive that speaking up with concerns or ideas to improve safety is valued and reinforced. The foregoing climates might not be as strong in acting on these suggestions, however, and clear structures that support accountability for patient harm may be lacking.

Fig 1.

Fig 1

Open in a new tab

The 5 safety climate profile shapes. Scores on each of the 10 dimensions have been standardized to demonstrate the relative relationships among these dimensions.

The third climate shape, a comprehensive climate shape, represents climates that are uniformly high across all dimensions. The team-oriented shape describes climates in which teamwork within the unit and across units is uniformly perceived as more positive than other dimensions. In these climates, team members rally among themselves in support of safety, but may not sense that they have significant support from local or organizational level leadership. Finally, conflicting climate shapes refer to those in which local leadership and frontline staff demonstrate and perceive a local commitment to patient safety, but might not perceive similar support from organizational level leaders or other units with which they work. Table 1 displays raw mean scores on each climate dimension by climate shape, as well as ICC and other descriptive information.

CLABSI

CLABSI data were collected during a 12-month baseline period before the start of the intervention for each cohort (2008 for cohort 1, 2009 for cohort 2 and a portion of cohort 1, 2008–2009 for cohort 3, and 2009–2010 for cohort 4). Specifically, the number of infections and the number of line-days (ie, number of days in which a central line is in place for all patients in a given ICU) for each unit were collected according to definitions provided by the CDC. Individual patient level data were not collected. The total number of CLABSIs was summed over the 12-month period to create a numerator for analysis, and line-days were summed over the same 12-month period for each unit to create a denominator for analysis. Details of infection rate data collection are reported elsewhere.25–27

Other covariates

Other covariates of interest included unit size (number of beds) and unit type (ie, specialty ICU, medical ICU, surgical ICU, or combined medical-surgical ICU). Unit bed size ranged from 4 beds to 50 beds (mean, 14.27; standard deviation [SD], 7.48). Specialty ICUs (n = 39) included burn, trauma, coronary, and specialized surgical ICUs that focused solely on cardiothoracic patients. In addition, traditional medical ICUs (n = 24) caring for adult and geriatric patients with complex medical needs, surgical ICUs (n = 13) caring for critically ill patients who have undergone complex surgical procedures, and combined medical-surgical ICUs (n = 161) that care for a wide range of patients with complex care needs were also included in the analyses.

Analyses

Descriptive summary statistics and descriptive graphs were created to explore variations in CLABSI rates, as well as the unadjusted relationships between the 3 safety climate profile characteristics and CLABSI rates. One ICU was dropped during data management processing for not reporting a valid denominator for their infection data (reported as 0); thus, analyses were conducted on 237 ICUs.

Exploratory data analyses including unadjusted histograms and stem-and-leaf plots revealed a large number of units reporting zero infections during the baseline period, suggesting that zero-inflated Poisson regression analyses should be used to fit observed data. Zero-inflated Poisson (ZIP) models assume that the series of zero outcomes may be predicted by different process than nonzero outcomes and thus include 2 separate models, 1 model examining the predictors of nonzero outcomes (ie, noninflated model) and 1 model examining zero outcomes (ie, inflated model).28,29 First, a series of unadjusted ZIP models were examined to investigate the crude relationship between the 3 safety climate profile characteristics and baseline CLABSI infection risks. These models were then extended to adjust for both unit size and type. All analyses were conducted with Stata/IC 12.1 for Windows (StataCorp, College Station, TX).

RESULTS

Relationships between the 3 safety climate profile characteristics and CLABSI

Overall, the mean CLABSI rate was 1.78 infections per 1000 central line-days (SD, 2.09) across all ICUs during the baseline period. Boxplots examining infection rates by unit type suggested meaningful variation among unit types, and lowess plots examining unadjusted relationships between unit size and rate also implicated size as a potential confounder of the safety climate–infection relationship.

We used an initial series of ZIP regression models to examine the unadjusted relationship between the three climate profile characteristics and infection risk. These models indicated that none of the profile characteristics were significantly related to the odds of having a unit infection rate of zero (Table 2). The noninflated portion of the model indicated that climate profile shape was significantly related to infection rates for units with infection rates greater than zero (Wald χ2 = 32.68; P < .0001). Specifically, the relative risk of infection was significantly higher in units with a conflicting climate shape (incident risk ratio [IRR], 1.70; P < .0001) and in units demonstrating a nonpunitive climate shape (IRR, 1.79; P < .0001) compared with units with a generative leadership shape. Profile elevation and profile variation were not significantly related to infection risk (IRR, 1.00; P = .19 and IRR, 0.99; P = .19, respectively).

Table 2.

Crude and adjusted zero-inflated Poisson regression coefficients

Crude Adjusted
Infections >0 Infections = 0 Infections >0 Infections = 0
Variable IRR 95% CI P value Coefficient 95% CI P value Coefficient 95% CI P value Coefficient 95% CI P value
Unit patient safety profile
  Profile elevation 1.00 0.97–1.02 .76 0.22 −0.11 to 0.55 .19 1.00 0.97–1.02 .74
Characteristics
  Profile variability 0.99 0.96–1.02 .51 0.22 −0.11 to 0.54 .19 0.99 0.96–1.02 .48
  Profile shape <.001w .26w <.001w
    Nonpunitive 1.79 1.33–2.38 <.001 −0.75 −5.55 to 4.05 .76 1.77 1.33–2.36 <.001
    Comprehensives 1.19 0.84–1.67 .33 2.93 −1.39 to 7.26 .18 1.19 0.85–1.69 .31
    Team-oriented 0.78 0.37–1.62 .51 0.7 −5.88 to 7.27 .84 0.61 0.30–1.23 .16
    Conflicting 1.7 1.35–2.13 <.001 1.3 −2.57 to 5.17 .51 1.57 1.25–1.98 <.001
Other unit
  Unit size 1.00 0.99–1.00 .28 −0.48 −0.48 .02
Characteristic
  Unit type <.001w
    Medical ICU 0.86 0.68–1.09 .22
    Medical-surgical ICU 0.78 0.66–0.93 <.01
    Surgical ICU 0.32 0.19–0.53 <.001
Open in a new tab

w, wald test

s, generative leadership is the referent climate profile shape.

*

Adjusted model includes unit size and unit type covariates, as well as an inflated factor for unit size.

To adjust for potential confounding variables and examine other unit characteristics likely to influence the odds of a unit having zero infections, we next examined a second series of ZIP models that included unit size and type. Three different models were explored: a model that fit unit size as a covariate in both the Poisson and inflated models, a model that fit both unit size and unit type as covariates in both the Poisson and inflated models, and a model that fit unit type and unit size as covariates in the Poisson model, but fit only unit size in the inflated model. Indices of model fit, including the Akaike information criterion, and qualitative investigation of model residual plots suggested that the model fitting unit type and unit size as covariates in the Poisson model but including only unit size in the inflated model provided the best fit.

The results summarized in Table 2 indicate that unit size was significantly related to the odds of a unit having zero infections (odds ratio, 0.62; P = .02). This indicates that larger ICUs were less likely to have zero infections and, specifically, that for each additional bed added to a unit, the odds of reporting zero infections were reduced by 38%. Unit type was significantly related to the relative risk for CLABSI (Wald χ2 = 22.86; P < .0001). The risk of infection in specialty ICUs (eg, burn, trauma) was 22% higher compared with medical-surgical ICUs (IRR, 0.78; P < .01) and 68% higher compared with surgical ICUs (IRR, 0.32; P < .001).

After adjusting for unit type and unit size, climate profile shape remained a significant predictor of infection risk (Wald χ2 = 36.63; P < .0001); however, profile elevation and profile variability again were not significantly related to infection risk (P = .74 and .48, respectively). These results indicate that the incidence rate of infection was 77% higher in units with a nonpunitive climate shape (IRR, 1.77; P < .001) and 57% higher in units with a conflicting climate shape (IRR, 1.57; P < .001) compared with units with the generative leadership shape. Generative leadership climate served as the reference group given that theoretically and empirically, it is the most positive, and also most balanced, climate shape; that is, climate scores were high on both unit-referenced dimensions like unit leader actions and support for safety and hospital-referenced dimensions like teamwork across units.

DISCUSSION

Understanding the role that organizational factors, such as patient safety climate, play in shaping clinician behavior and patient outcomes is critical for improving the quality and safety of care provided to some of the most at-risk patients. Our findings indicate that patient safety climate, when operationalized in terms of climate profile characteristics, is significantly related to the CLABSI rate in ICUs after controlling for other unit factors, such as size and type. Although simple summary profile characteristics, such as profile elevation and profile variation, were not meaningfully related to incidence rates, climate profile shape was significantly related to incidence rates above zero. This finding suggests that the various aspects of safety climate are neither additive nor interchangeable, and that it is the constellation of factors that compose the safety climate, as well as the complex patterns and interactions among them, that must be studied to understand the relationship between patient safety climate and patient outcomes. Our findings also align with previous preliminary work across a range of acute care areas in which profile elevation and variation were related to patient ratings of their care experience, but only profile shape was related to the number of adverse events occurring in the unit.33

Our results indicate that ICUs with conflicting climates and nonpunitive climates demonstrated greater relative CLABSI risk compared with ICUs with a generative leadership climate, but that climate shape is not related to the odds of a unit reporting zero infections. Whereas climate profile shape was found to be related to infection rates only in those ICUs reporting rates above zero, these findings underscore the importance of safety climate in units that continue to struggle with sustaining a zero rate over time. Theoretically, these findings align with previous work highlighting that a climate of safety and sustainment of safe outcomes is the product of multiple interacting factors, including leadership actions that emphasize and reinforce safety, proactive identification of potential threats to safety, and continuous learning.34 They also supplement previous findings indicating that individual climate domains that pertain to teamwork both within and across work areas, as well as leadership, are related to composite indices of the AHRQ patient safety indicators.12 Conversely, units with a conflicting climate shape or a nonpunitive climate shape demonstrated significantly higher infection risks.

The theoretical pathways underlying these findings may differ in meaningful ways, however; for example, conflicting climates were characterized by discrepancies between hospital-level climate domains and unit-level climate domains. This suggests that in these units, team members may perceive that local leaders and direct colleagues support and value safety, but do not perceive the same level of support for safety from hospital level leaders or other areas of the organization. This theoretically could reflect conflicting or unclear goals that may tacitly reinforce unsafe behavior (eg, using workarounds or shortcuts to improve efficiency). For units characterized by nonpunitive climates, the pathway may look different. Documented reporting bias and underreporting35–37 of hospital-acquired harm has led to efforts to create nonpunitive, psychologically safe climates in which clinicians and staff feel comfortable speaking up about potential or reals harms that they observe or encounter.38 Team members working in units characterized by a nonpunitive climate may be more likely to identify and discuss real or potential hazards, as well as to more accurately document instances of hospital-acquired infection. This is an important step on the road to improvement and enhances the validity of their rate data, but may make it more difficult to detect true relationships between other climate shapes and rates, given that some underreporting may still occur in units characterized by other climate shapes.

Our findings must be considered in light of several study limitations. First, this study was secondary analysis of a subset of a larger intervention study. There is the potential that ICUs included in these analyses are not entirely representative of all ICUs, though they did represent both teaching and nonteaching hospitals, as well as rural and nonrural settings. In addition, their safety climate scores were in line with national benchmarks reported by the AHRQ based on their national HSOPS database, and the average infection rate across included ICUs was in line with national averages reported by the CDC. Second, infection incidence rates were reported at the unit level of analysis, and there was no direct way to adjust for patient case mix. Unit type was included as a proxy measure for patient acuity, and models were adjusted for this. Moreover, directionality or causality cannot be inferred from these analyses. Another important potential limitation includes other factors that we were not able to control for in these analyses that might have impacted or confounded findings. For example, units included in these analyses might have previously implemented interventions outside of this project to improve CLABSI rates, the safety culture, or both. Finally, we evaluated the relationship between safety climate and baseline CLABSI rate. The relationship between climate and CLABSI might have been different had we evaluated improvement in CLABSI or postintervention CLABSI rate.

Overall, our results suggest that the relative risk of CLABSI in US adult ICUs is related to patient safety climate profile shape. In addition, they underscore the value of understanding and operationalizing patient safety climate, in line with theoretical definitions that define climate as the pattern among dimensions, rather than examining the effects of individual dimensions of climate in a piecemeal manner. Finally, these findings suggest that hospital leaders, managers, clinicians, and staff should work to foster a safety climate characterized by high levels of both local unit leader and hospital leader commitment to supporting their clinicians and staff in their efforts to optimize patient safety, as well as high levels of teamwork both within and across units.

Acknowledgment

We thankfully recognize the contributions of other members of the multi-disciplinary team that lead the National Stop BSI project: Sean Berenholtz, Christine Goeschel, David Thompson, Lisa Lubomski, Marianna Lesher, Justin St. Andre, Stephen Hines, and Sam Watson, as well as many others brought this work to fruition. We are also grateful to the ICU teams participating in the Stop BSI project for their time and dedication to this work. Finally, we are appreciative of the constructive comments and the feedback from3 reviewers that helped refine this manuscript.

The original Stop BSI work was supported in part by funding from the Agency for Healthcare Research and Quality (AHRQ), (Contract HHSA290200600022; Task Order 7). The secondary analyses and work reported here was supported by the Johns Hopkins Institute for Clinical and Translational Research (ICTR), which is funded in part by Grant 1KL2TR001077-01 from the National Center for Advancing Translational Sciences (NCATS), a component of the National Institutes of Health (NIH), and NIH Roadmap for Medical Research. The contents are solely the responsibility of the authors and do not necessarily reflect the official views of AHRQ, ICTR, NCATS, or NIH.

Publication of this article was supported by the Agency for Healthcare Research and Quality (AHRQ).

Footnotes

Conflicts of interest: None to report

References

  • 1.Centers for Disease Control and Prevention. Making healthcare safer: reducing bloodstream infections. [Accessed June 10, 2013];CDC Vital Signs. 2011 Available from: http://www.cdc.gov/VitalSigns/pdf/2011-03-vitalsigns.pdf.
  • 2.Scott RD. The direct medical costs of healthcare-associated infections in US hospitals and the benefits of prevention. Atlanta [GA]: 2009. pp. 5–12. [Google Scholar]
  • 3.Pronovost PJ, Weast B, Rosenstein B, Sexton JB, Holzmueller CG, Paine L, et al. Implementing and validating a comprehensive unit-based safety program. J Patient Saf. 2005;1:33–40. [Google Scholar]
  • 4.Marsteller JA, Sexton JB, Hsu Y-J, et al. A multicenter, phased, cluster-randomized controlled trial to reduce central line-associated bloodstream infections in intensive care units. Crit Care Med. 2012;40:2933–2939. doi: 10.1097/CCM.0b013e31825fd4d8. [DOI] [PubMed] [Google Scholar]
  • 5.Zohar D, Luria G. A multilevel model of safety climate: cross-level relationships between organization and group-level climates. J Appl Psychol. 2005;90:616–628. doi: 10.1037/0021-9010.90.4.616. [DOI] [PubMed] [Google Scholar]
  • 6.Weaver SJ, Lubomksi LH, Wilson RF, Pfoh ER, Martinez KA, Dy SM. Promoting a culture of safety as a patient safety strategy: a systematic review. Ann Intern Med. 2013;158(5 Pt 2):369–374. doi: 10.7326/0003-4819-158-5-201303051-00002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Landrigan CP, Parry GJ, Bones CB, Hackbarth AD, Goldmann DA, Sharek PJ. Temporal trends in rates of patient harm resulting from medical care. N Engl J Med. 2010;363:2124–2134. doi: 10.1056/NEJMsa1004404. [DOI] [PubMed] [Google Scholar]
  • 8.Leape LL, Berwick DM, Bates DW. What practices will most improve safety? Evidence-based medicine meets patient safety. JAMA. 2002;288:501–507. doi: 10.1001/jama.288.4.501. [DOI] [PubMed] [Google Scholar]
  • 9.Leape LL, Berwick DM. Five years after to Err Is Human: what have we learned? JAMA. 2005;293:2384–2390. doi: 10.1001/jama.293.19.2384. [DOI] [PubMed] [Google Scholar]
  • 10.Huang DT, Clermont G, Kong L, et al. Intensive care unit safety culture and outcomes: a US multicenter study. Int J Qual Health Care. 2010;22:151–161. doi: 10.1093/intqhc/mzq017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Shortell SM, Zimmerman JE, Rousseau DM, et al. The performance of intensive care units: does good management make a difference? Med Care. 1994;32:508–525. doi: 10.1097/00005650-199405000-00009. [DOI] [PubMed] [Google Scholar]
  • 12.Mardon RE, Khanna K, Sorra J, Dyer N, Famolaro T. Exploring relationships between hospital patient safety culture and adverse events. J Patient Saf. 2010;6:226–232. doi: 10.1097/PTS.0b013e3181fd1a00. [DOI] [PubMed] [Google Scholar]
  • 13.Gershon RR, Karkashian CD, Grosch JW, et al. Hospital safety climate and its relationship with safe work practices and workplace exposure incidents. Am J Infect Control. 2000;28:211–221. doi: 10.1067/mic.2000.105288. [DOI] [PubMed] [Google Scholar]
  • 14.MacDavitt K, Chou S-S, Stone PW. Organizational climate and health care outcomes. Jt Comm J Qual Patient Saf. 2007;33(11) Suppl:45–56. doi: 10.1016/s1553-7250(07)33112-7. [DOI] [PubMed] [Google Scholar]
  • 15.Rosen AK, Singer S, Zhao S, Shokeen P, Meterko M, Gaba D. Hospital safety climate and safety outcomes: is there a relationship in the VA? Med Care Res Rev. 2010;67:590–608. doi: 10.1177/1077558709356703. [DOI] [PubMed] [Google Scholar]
  • 16.Singer SJ, Vogus TJ. Safety climate research: taking stock and looking forward. BMJ Qual Saf. 2013;22:1–4. doi: 10.1136/bmjqs-2012-001572. [DOI] [PubMed] [Google Scholar]
  • 17.Carr JZ, Schmidt AM, Ford JK, DeShon RP. Climate perceptions matter: a meta-analytic path analysis relating molar climate, cognitive and affective states, and individual-level work outcomes. J Appl Psychol. 2003;88:605–619. doi: 10.1037/0021-9010.88.4.605. [DOI] [PubMed] [Google Scholar]
  • 18.Meyer AD, Tsui AS, Hinings CR. Configural approaches to organizational analysis. Acad Manag J. 1993;36:1175–1195. [Google Scholar]
  • 19.Schulte M, Ostroff C, Shmulyian S, Kinicki A. Organizational climate configurations: relationships to collective attitudes, customer satisfaction, and financial performance. J Appl Psychol. 2009;94:618–634. doi: 10.1037/a0014365. [DOI] [PubMed] [Google Scholar]
  • 20.Lipitz-Snyderman A, Steinwachs D, Needham DM, Colantuoni E, Morlock LL, Pronovost PJ. Impact of a statewide intensive care unit quality improvement initiative on hospital mortality and length of stay: retrospective comparative analysis. BMJ. 2011;342:d219. doi: 10.1136/bmj.d219. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Berenholtz SM, Pham JC, Thompson DA, et al. Collaborative cohort study of an intervention to reduce ventilator-associated pneumonia in the intensive care unit. Infect Control Hosp Epidemiol. 2011;32:305–314. doi: 10.1086/658938. [DOI] [PubMed] [Google Scholar]
  • 22.Sorra JS, Nieva VF. Prepared by Westat, under Contract No. 290-96-0004. Rockville, MD: Agency for Healthcare Research and Quality; 2004. Hospital Survey on Patient Safety Culture. [Google Scholar]
  • 23.Sorra JS, Dyer N. Multilevel psychometric properties of the AHRQ hospital survey on patient safety culture. BMC Health Serv Res. 2010;10:199. doi: 10.1186/1472-6963-10-199. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Real K, Kim S, Conigliaro J. Using a validated health promotion tool to improve patient safety and increase health care personnel influenza vaccination rates. Am J Infect Control. 2013;41:691–696. doi: 10.1016/j.ajic.2012.09.027. [DOI] [PubMed] [Google Scholar]
  • 25.Health Research and Educational TrustJohns Hopkins Medicine Armstrong Institute for Patient Safety and Quality, Michigan Health and Hospital Association Keystone Center for Patient Safety and Quality. Eliminating CLABSI, a national patient safety imperative: final report on the national On the CUSP: Stop BSI project. Washington [DC]: Agency for Healthcare Research and Quality; 2012. Publication 12-0087-EF. [Google Scholar]
  • 26.Lin DM, Weeks K, Bauer L, et al. Eradicating central line-associated bloodstream infections statewide: the Hawaii experience. Am J Med Qual. 2012;27:124–129. doi: 10.1177/1062860611414299. [DOI] [PubMed] [Google Scholar]
  • 27.Hong AL, Sawyer MD, Shore A, et al. Decreasing central line–associated bloodstream infections in Connecticut intensive care units. J Healthc Qual. 2013;35:78–87. doi: 10.1111/j.1945-1474.2012.00210.x. [DOI] [PubMed] [Google Scholar]
  • 28.Atkins DC, Gallop RJ. Rethinking how family researchers model infrequent outcomes: a tutorial on count regression and zero-inflated models. J Fam Psychol. 2007;21:726–735. doi: 10.1037/0893-3200.21.4.726. [DOI] [PubMed] [Google Scholar]
  • 29.Loeys T, Moerkerke B, De Smet O, Buysse A. The analysis of zero-inflated count data: beyond zero-inflated Poisson regression. Br J Math Stat Psychol. 2012;65:163–180. doi: 10.1111/j.2044-8317.2011.02031.x. [DOI] [PubMed] [Google Scholar]
  • 30.Calinski T, Harabasz J. A dendrite method for cluster analysis. Commun Stat. 1974;3:1–27. [Google Scholar]
  • 31.Cameron KS, Quinn RE. Diagnosing and changing organizational culture based on the competing values framework. 3rd ed. San Francisco [CA]: Jossey-Bass; 2011. [Google Scholar]
  • 32.Zohar D, Livne Y, Tenne-Gazit O, Admi H, Donchin Y. Healthcare climate: a framework for measuring and improving patient safety. Crit Care Med. 2007;35:1312–1317. doi: 10.1097/01.CCM.0000262404.10203.C9. [DOI] [PubMed] [Google Scholar]
  • 33.Weaver SJ. A configural approach to patient safety climate: the relationship between climate profile characteristics and patient outcomes [dissertation] Orlando, FL: University of Central Florida; 2011. [Google Scholar]
  • 34.Vogus TJ, Sutcliffe KM, Weick KE. Doing no harm: enabling, enacting, and elaborating a culture of safety in health care. Acad Manag Perspect. 2010;24:60–77. [Google Scholar]
  • 35.Fontela PS, Quach C, Buckeridge D, Pai M, Platt RW. Surveillance length and validity of benchmarks for central line–associated bloodstream infection incidence rates in intensive care units. PLoS ONE. 2012;7:e36582. doi: 10.1371/journal.pone.0036582. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Backman LA, Melchreit R, Rodriguez R. Validation of the surveillance and reporting of central line–associated bloodstream infection data to a state health department. Am J Infect Control. 2010;38:832–838. doi: 10.1016/j.ajic.2010.05.016. [DOI] [PubMed] [Google Scholar]
  • 37.Sexton DJ, Chen LF, Moehring R, Thacker PA, Anderson DJ. Casablanca redux: we are shocked that public reporting of rates of central line–associated bloodstream infections are inaccurate. Infect Control Hosp Epidemiol. 2012;33:932–935. doi: 10.1086/667383. [DOI] [PubMed] [Google Scholar]
  • 38.Nembhard IM, Edmondson AC. Making it safe: the effects of leader inclusiveness and professional status on psychological safety and improvement efforts in health care teams. J Organ Behav. 2006;27:941–966. [Google Scholar]

RESOURCES